Hugging Face Papers ↗ · 2026-11-18
由编程智能体对光学神经网络进行原位训练
媒体报道 Hugging Face Papers 原文语言:中文
编程智能体通过使用可解释的光学元件对光学神经网络进行重新参数化,仅凭测量数据对其进行训练。该方法实现了比无梯度方法更好的分类和图像编码效果,且以较低的成本接近了通过校准模型进行反向传播的性能。
全文
A coding agent trains an optical neural network from measurements alone by reparametrizing it with interpretable optical elements, achieving better classification and image encoding than gradient-free methods and approaching backpropagation through a calibrated model at lower cost.
正文以来源语言显示。